This paper identifies and interprets the six essential steps in the develop
ment of neoclassical growth theory from the Solow-Swan model to the new gro
wth theory models of sustained endogenous growth driven by learning (e.g.,
[Romer, P., 1990. Endogenous technological change. Journal of Political Eco
nomy, 98 (5): S71-S102]). The success of that modeling exercise is based on
the specification of selected assumptions about learning in industry. The
practical shortcomings of the exercise are highlighted by the various and c
onflicting policy implications arising from the selection of attributes. It
is concluded that better policies may follow from basing assumptions of le
arning behavior on the understanding of the process and strategy of learnin
g that is emerging from institutional and evolutionary work. (C) 2000 Elsev
ier Science B.V. All rights reserved.